I have spent way too long faffed around with my own dust cyclone design, altered and re-altered it, and been totally frustrated by the lack of clear, intelligible, accessible information on the subject. So I'm dropping a few notes for you all out there in the hope that other people will not have to suffer through the long and frustrating design process that I went through.
First then, dust cyclone designs very a lot - from 1D3D, 2D2D, Bill Pentz 1.64 cone sizes, etc. Looking also at the inlet, you will see papers detailing a number of different sizes and layouts for blasting the air in. From this, I initially and foolishly assumed there was a lot of design flexibility when it came to making my own cyclone - I was wrong!! Best I can tell, each design is made, and then altered and tuned to make it function correctly. Just throwing together a design with "about this dimension" numbers in it, is a recipe for failure.
So First... Maybe... Just dont?
Simple and up front - unless you are dedicated to learning all about cyclones, or want to make something than you just cannot find already on the shelf, it is so much easier to just buy one. And if you do have to make one, go with a tried and tested design/shape. Bill Pentz has a spreadsheet for calculating here and sells his own brand of large, workshop scale cyclones here or see link below. When you buy a professional solution like this, you can be damn sure it is well designed and will do its job.
Honestly, I wish I knew you could just buy these cyclones before I attempted to make myself one - small, simple cyclones are really pretty cheap. Of course, if you want to do it for literally nothing, you can just get a random bin, shove an exit port out the top, add an inlet port to the side, and it will do a good job of grabbing larger bits, but not small particulates. From a health point of view, these are the dangerous one. The small, simple off the shelf cones claim to filter above 10 microns, which is pretty good if you believe them and a great improvement on a bin with holes in it. You can always add your own HEPA afterwards if designed correctly. Whereas if you want to make your own cyclone, you can easily end up spending more on materials than the cost to simply buy an existing solution, not to mention the time and heachache of it all - and the end result probably wont be as good.
Also note, that if you are running with a high flow rate, you can put multiple cyclones in parallel, dropping into the same or different bins. Of course, larger outlets are less prone to clogging if you are expecting larger pieces to be flying through.
Cyclone Design Considerations
So... you are still here? Well OK! Disclaimer though, I am still a long way from an expert on any of this, more just a layman trying to help other people out. That said.....
Here are some core concepts and thoughts that I learned in part from experiment, and part from mailing Bill Pentz. I should add before I go any further on, dont mail Bill Pentz unless you really, really need to. In fact, even then think twice. I'm serious. Just from the brief email conversation I had with him, the guy gets a truly astonishing number of e-mails, like mind bogglingly so, and I get the distinct impression that they stress him out no small amount - so give the guy some space? In fact, part the reason I am writing this is to give out information is so that perhaps, he wont need to. So then, core principles, design issues, etc...
Cone Size and How it Works
It looks a lot to me that, for a given design, there is a certain length that the cyclone "wants" to be. If the cone is too long, the air flow will separate from its correct path, reverse too early and it will not clear. Material will get stuck in there, spinning round and round and not being able to exit the chute at the bottom. If the cone is too short, then the cyclone extends down into the bin and sucks dust up. If you have a design that is too long (not clearing) then I was told you can reduce the height by cutting off small sections from the bottom until you find the correct length - but it might screw up your air flow a bit in other ways.
Think also about the way the air is moving through, and how the effect works. The centrifugal force pushes material to the outside of the body and the cone, then it spins down towards the bottom. At the bottom, the direction of the air reverses, and the dust and particulates - now at the edge of the flow, and so the lowest point when direction reverses - attempt to follow their existing path, and are kicked out and into the relatively still air of the bin. As such, the more turbulence there is in the flow of the cyclone, the larger the particulates that will be pulled out from the edge of the flow, and so not be in position to be removed as air moves into the inner spiral. I havent included a picture of this process, there are loads out there - just google them, I dont want to steal!
Essentially though, unless you are using an existing and scalable plan like Pentz's, there is no magic recipe. It is a complex system, and I suspect that industrial designs go through considerable R&D involving finite modelling and multiple iterations to get the dimensions just right. So too, unless you get lucky you might also have to go through this process.
Lots of air and bits swirling at speed inside an ungrounded, insulated plastic cone means static electricity. I made mine out of a large clear acrylic tube and a 3d printed cone painted in super thin epoxy to harden it up. A lot of material was sticking to the edges. I fixed this with some aluminium foil stuck to the inside of the cone on one side, grounded. For the earth connection on the foil, I stuck a piece of wire underneath it and glued it down. To make the glue, I ground the edge off a piece of mild steel (magnet under grinding wheel to catch the filings), mixed with epoxy, and then used a magnet when in place to make the filings align and create an earth connection. Not my idea alas, I cant take the credit! I also used this glue to connect multiple pieces of alu foil together, although just grounding one side of the cone pretty much sorted it. However, this process was a bit long winded, and I understand aluminium tape should also work.
A side note, some plastics will not hold static electricity - see the Oneida cone above, designed to be anti-static.
Motor / Impeller
For my design, I wanted a super quiet extractor. I liked the idea of the Dust Sniper, which is essentially a couple of vacuum cleaners in a sound proofed cabinet drawing through two dust cyclones. Personally however I didnt see the point in using old vacuum cleaners when you can buy vacuum cleaner motor / impellers cheap online, and design a cabinet specifically for them. Vacuum cleaners have a number of design considerations (weight, movability, size) that dont come into play with stationary workshop extractors. Of course, just slapping a cyclone on the front end of a vaccum cleaner is by far athe easiest for most people unless you have specific other needs as I did. In my case, I needed more extraction than a single vacuum could provide, I wanted to be able to draw from the same inlet (not a good idea to use two different vacuums in parallel like this, as we shall see), and I wanted it to be super quiet. So I bought my own motor/impellers, and build a cabinet just for them.
Note to US readers - you might have to shop around to find a good unit that plays nicely with 110V. Looking on Alibaba there are a few that are rated for both 110V and 220V, but with reduced power at the lower voltage. My suspicion is that most the cheap motors you can find online that are rated to 220V will run 110V, but will lower power - however you will need do you own research if you want to try this!
More generally for dust extractors, people typically stay away from the vacuum cleaner motors as they produce *relatively* high pressure, lower flow. Larger impellers will be the reverse - relatively low pressure, high flow. I personally discarded this for two reasons. Firstly because I wanted to push my air through a few different stages including a HEPA filter, which typically have a fairly large pressure drop over them (at least for concertina style HEPA), and secondly because of sourcing. To buy a good sized impeller which was up to the task didnt look easy or cheap. On the other hand, when I checked youtube for vids of people making them - well, it just looked like a total ball ache. So I whacked two vacuum cleaner motors in parallel, and used an AC Motor Speed Controller to alter the flow rate.
A note though, apparently running more than one motor can be a bit dodgy, especially if your ducting is not wide enough or if you use two differently sized motors. Restrict the flow, and your motors will be pulling against too much pressure, overheat, and burn down your house. Use motors that are not the same, and one will get all that nice cooling air juice, while the other will be restricted, sit there, and explode or again, burn down your house. Alternatively, I have been informed that two motors in series will actually produce close to twice the flow, but they will obviously be spinning much faster and so drawing more current. As such overheating is more of an issue - bringing us back to the burning down the house and dying thing. Or so I am told. In other words, dont sue me if you screw up and die. That said, I'm happy with the power of my setup in terms of suction, and comfortable it is not going to kill me. My motors are identical and running in parallel, and I've been checking the power usage on each with a wattmeter to make sure nothing is getting screwy. As implied, vacuum cleaner motors cool themselves using the air the impellers are blowing, it is literally vented past the motor windings. As such it is a little bit important that the air passing through is nice and clean. Dont want dust building up in there... (see burning/dying)
If you do want to make your own impeller, you'll need to be careful it is both strong enough for the RPM and balanced. You "might" be able to 3d print something - there are designs on thingiverse although I have not played with them. Fibre reinforced plastic might be a good idea, but with the size and scale you'd be running at, that plastic is going to be under a fair bit of stress. Alternatively, a CNC router could help. Design with straight veins to simplify, and use good quality plywood. Spin it to find any imbalance, and remove material from that side to get it true. That said, I didnt go down this route myself so... you're on your own!
I hope this article has been helpful. Hopefully it will save some of you the total headache I have been though. And with that said - good luck!!